From chris at qwirx.com Thu Apr 1 17:11:26 2010 From: chris at qwirx.com (Chris Wilson) Date: Thu, 1 Apr 2010 18:11:26 +0200 (CEST) Subject: [Box Backup] Feature request - for lousy connections In-Reply-To: <4BB21C58.1020009@orbie.de> References: <4BB1F7D0.2090609@orbie.de> <4BB20673.7080300@orbie.de> <4f5fd957e5aa50e2813ec254360a9b06@localhost> <4BB215A4.4080905@orbie.de> <4BB21C58.1020009@orbie.de> Message-ID: Hi Florian, On Tue, 30 Mar 2010, Florian Eyben wrote: > My question was, if you think it is possible to save the state when the > connection drops (i.e. save the last state before the transfer of the > file or the action where the connection dropped). I mean, more or less, > call the function that saves the state (currently when stopping the > daemon), when an error occurs. The only significant state that we could save would be the directory listings. It should be possible to do this, but I'd like to understand what the real issue is that you're experiencing, and whether this is the best way to fix it, before actually implementing it. Otherwise it's not getting you any closer to a solution, and taking away time that I could spend on implementing more useful new features. Cheers, Chris. -- _ ___ __ _ / __/ / ,__(_)_ | Chris Wilson <0000 at qwirx.com> - Cambs UK | / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Perl/SQL/HTML Developer | \ _/_/_/_//_/___/ | We are GNU-free your mind-and your software | From chris at qwirx.com Thu Apr 1 17:15:07 2010 From: chris at qwirx.com (Chris Wilson) Date: Thu, 1 Apr 2010 18:15:07 +0200 (CEST) Subject: [Box Backup] Mass Deletion Protection In-Reply-To: <74d01c7a1003300637y415e415s89538eb8bfd22211@mail.gmail.com> References: <01b8d069eaf052417b941e26c9f33047@localhost> <7bb0a901cafd6f578852158e3e86b2c5@localhost> <74d01c7a1003300637y415e415s89538eb8bfd22211@mail.gmail.com> Message-ID: Hi Pete, On Tue, 30 Mar 2010, Peter Jalajas, GigaLock Backup Services wrote: >> Pete's suggestion of using "list -otrsd" seems to do the trick: the >> data is still there, I just was not able to look for it in the right >> way. > > One thing I just noticed recently, and, Chris, maybe it's new behavior, > is that: 1) the -o option shows old _and_ current versions (and no > deleted versions), but 2) the -d option shows _only_ the deleted > versions (and no old or current versions). > > I thought I recalled that I had previously routinely used -odstr to list > all my files and then used grep to find the ones I was looking for. > So, I believe now there is no way to use a single command to list _all_ > versions of a file. Chris: what do you think? It's not 100% clear to me what you mean, but you seem to be saying that if you use "list -od", you don't see both old and deleted versions of files? Please could you demonstrate? I would expect that you would see both, and the code certainly seems to support that. If both options are given, and unless there is a subtle bug in the code, no files are excluded from the directory listing at all. Cheers, Chris. -- _ ___ __ _ / __/ / ,__(_)_ | Chris Wilson <0000 at qwirx.com> - Cambs UK | / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Perl/SQL/HTML Developer | \ _/_/_/_//_/___/ | We are GNU-free your mind-and your software | From joseyluis at gmail.com Tue Apr 6 03:36:21 2010 From: joseyluis at gmail.com (Jose Luis Rodriguez Garcia) Date: Tue, 6 Apr 2010 04:36:21 +0200 Subject: [Box Backup] Patch for raidfile test on NetBSD 4/5 Message-ID: 1- I have created a patch for raidfile test. I copy the patch at the end of the email. If I must create a trac ticket, please create a account trac. Login: joseyluis With this patch NetBSD pass the raidfile test. I have tested it on NetBSD 4. The problem is that in NetBSD 4/5 the readdir and openddir functions are called in libc: __readdir30 and __closedir30. readdir and closedir exist for compatibility with NetBSD 3 and before. There was a change in the ABI between the 3 and 4 version for enable 64 bits types. 2- What is the purpose of intercept these function calls in the intercept.cpp file? --- intercept.h.orig 2010-04-04 01:58:54.000000000 +0200 +++ intercept.h 2010-04-04 02:00:33.000000000 +0200 @@ -13,6 +13,18 @@ #include +#ifdef __NetBSD__ //__NetBSD_Version__ is defined in sys/param.h +#include +#endif + +#if defined __NetBSD_Version__ && __NetBSD_Version__ >= 399000800 //3.99.8 vers. +#define FUNC_OPENDIR "__opendir30" +#define FUNC_READDIR "__readdir30" +#else +#define FUNC_OPENDIR "opendir" +#define FUNC_READDIR "readdir" +#endif + #include #include --- intercept.cpp.orig 2010-04-04 01:58:54.000000000 +0200 +++ intercept.cpp 2010-04-04 02:00:34.000000000 +0200 @@ -514,7 +514,7 @@ { if (opendir_real == NULL) { - opendir_real = (opendir_t*)find_function("opendir"); + opendir_real = (opendir_t*)find_function(FUNC_OPENDIR); } if (opendir_real == NULL) @@ -547,7 +547,7 @@ if (readdir_real == NULL) { - readdir_real = (readdir_t*)find_function("readdir"); + readdir_real = (readdir_t*)find_function(FUNC_READDIR); } if (readdir_real == NULL) From joseyluis at gmail.com Tue Apr 6 17:49:34 2010 From: joseyluis at gmail.com (Jose Luis Rodriguez Garcia) Date: Tue, 6 Apr 2010 18:49:34 +0200 Subject: [Box Backup] Patch for raidfile test on NetBSD 4/5 In-Reply-To: References: Message-ID: This patch also fixes the bbackupd test. On Tue, Apr 6, 2010 at 4:36 AM, Jose Luis Rodriguez Garcia wrote: > 1- I have created a patch for raidfile test. I copy the patch at the > end of the email. > > If I must create a trac ticket, please create a account trac. Login: joseyluis > > With this patch NetBSD pass the raidfile test. I have tested it on NetBSD 4. > > The problem is that in NetBSD 4/5 the readdir and openddir functions > are called in libc: __readdir30 and __closedir30. > readdir and closedir exist for compatibility with NetBSD 3 and before. > > There was a change in the ABI between the 3 and 4 version for enable > 64 bits types. > > 2- What is the purpose of intercept these function calls in the > intercept.cpp file? > > > --- intercept.h.orig 2010-04-04 01:58:54.000000000 +0200 > +++ intercept.h 2010-04-04 02:00:33.000000000 +0200 > @@ -13,6 +13,18 @@ > > #include > > +#ifdef __NetBSD__ //__NetBSD_Version__ is defined in sys/param.h > +#include > +#endif > + > +#if defined __NetBSD_Version__ && __NetBSD_Version__ >= 399000800 > //3.99.8 vers. > +#define FUNC_OPENDIR "__opendir30" > +#define FUNC_READDIR "__readdir30" > +#else > +#define FUNC_OPENDIR "opendir" > +#define FUNC_READDIR "readdir" > +#endif > + > #include > #include > > --- intercept.cpp.orig 2010-04-04 01:58:54.000000000 +0200 > +++ intercept.cpp 2010-04-04 02:00:34.000000000 +0200 > @@ -514,7 +514,7 @@ > { > if (opendir_real == NULL) > { > - opendir_real = (opendir_t*)find_function("opendir"); > + opendir_real = (opendir_t*)find_function(FUNC_OPENDIR); > } > > if (opendir_real == NULL) > @@ -547,7 +547,7 @@ > > if (readdir_real == NULL) > { > - readdir_real = (readdir_t*)find_function("readdir"); > + readdir_real = (readdir_t*)find_function(FUNC_READDIR); > } > > if (readdir_real == NULL) > From chris at qwirx.com Tue Apr 6 21:55:19 2010 From: chris at qwirx.com (Chris Wilson) Date: Tue, 6 Apr 2010 22:55:19 +0200 (CEST) Subject: [Box Backup] Patch for raidfile test on NetBSD 4/5 In-Reply-To: References: Message-ID: Hi Jose, On Tue, 6 Apr 2010, Jose Luis Rodriguez Garcia wrote: > 1- I have created a patch for raidfile test. I copy the patch at the > end of the email. Thanks for the patch! Please do post it on our bug tracker to ensure that it doesn't get forgotten. I will need to review it to ensure that it meets our coding style and doesn't break other platforms. Will you continue to test future versions of Box Backup on NetBSD 4/5? Would you mind being the official (or unofficial) maintainer of support for this platform? > The problem is that in NetBSD 4/5 the readdir and openddir functions > are called in libc: __readdir30 and __closedir30. > readdir and closedir exist for compatibility with NetBSD 3 and before. > > There was a change in the ABI between the 3 and 4 version for enable > 64 bits types. Perhaps we need to use the 64-bit versions instead? I guess that these, and not the 32-bit versions, are called by the standard readdir() and closedir() functions in C code? > 2- What is the purpose of intercept these function calls in the > intercept.cpp file? To simulate read errors and weird conditions (e.g. very large directories with tens of thousands of files), in order to test that Box Backup copes with them properly, without actually creating those conditions on disk. Cheers, Chris. -- _ ___ __ _ / __/ / ,__(_)_ | Chris Wilson <0000 at qwirx.com> - Cambs UK | / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Perl/SQL/HTML Developer | \ _/_/_/_//_/___/ | We are GNU-free your mind-and your software | From achim+box at qustodium.net Thu Apr 8 15:07:33 2010 From: achim+box at qustodium.net (Achim) Date: Thu, 08 Apr 2010 16:07:33 +0200 Subject: [Box Backup] Logging Message-ID: <01c375e6352797b3ab14e3f524ce6590@localhost> Hello list: A couple of remarks/questions/bugs on the logging front, in no particular order. Any insights/ideas welcome! 1) Duplicate notifications There was a discussion on the list a couple of years ago on the topic of surpressing duplicate notification (or not). I wanted to suggest that at least for the read-error, backup-error and store-full error, this should be configurable: "bang on if the user cannot be bothered to fix the problem". I think that in the current default configuration with surpression, scenario A) is more likely, but B) is a lot more desirable: A) problem happens, error message is sent, problem happens again, error message is surpressed, user thinks that problems went away B) problem happens, error message is sent, problem happens again, error message is sent, user is reminded about problem 2) I specified the logfile in bbackupd.conf as follows: LogAllFileAccess = yes LogFile = d:\test\log\bbackupd.log LogFileLevel = notice However, bbackupd.log is emptied at random intervals. Is this controllable? Why doesn't bbackupd always append to the LogFile, but decides to empty it? 3) the following error message (see last row) seems to be incorrect: "Can't open a file -- attempted to load a non-existant config file or bad file referenced within?" when the file is really locked by another process: this has nothing to do with a config file: 2010-04-08 15:33:49 [WARNING] Exception thrown: ConnectionException(Conn_SocketNameLookupError) at Socket.cpp(137) 2010-04-08 15:35:19 [WARNING] Failed to open file: D:\docs\profiles\firefox.new\parent.lock: El proceso no tiene acceso al archivo porque est siendo utilizado por otro proceso. (32) 2010-04-08 15:35:19 [WARNING] Exception thrown: CommonException(OSFileOpenError) at FileStream.cpp(144) 2010-04-08 15:35:19 [ERROR] Failed to upload file: D:\docs\profiles\firefox.new\parent.lock: caught exception: Common OSFileOpenError (Can't open a file -- attempted to load a non-existant config file or bad file referenced within?) (1/2) From chris at qwirx.com Thu Apr 8 15:25:18 2010 From: chris at qwirx.com (Chris Wilson) Date: Thu, 8 Apr 2010 16:25:18 +0200 (CEST) Subject: [Box Backup] Logging In-Reply-To: <01c375e6352797b3ab14e3f524ce6590@localhost> References: <01c375e6352797b3ab14e3f524ce6590@localhost> Message-ID: Hi Achim, On Thu, 8 Apr 2010, Achim wrote: > There was a discussion on the list a couple of years ago > on > the topic of surpressing duplicate notification (or not). I wanted to > suggest that at least for the read-error, backup-error and store-full > error, this should be configurable It is configurable: set NotifyAlways = yes if you want it. > "bang on if the user cannot be bothered to fix the problem". I think I agree that it should be on by default. > 2) I specified the logfile in bbackupd.conf as follows: > LogAllFileAccess = yes > LogFile = d:\test\log\bbackupd.log > LogFileLevel = notice > However, bbackupd.log is emptied at random intervals. Is this > controllable? Why doesn't bbackupd always append to the LogFile, but > decides to empty it? There was a bug where the file was emptied at the start of every run, but only on Windows, because the file was being opened in overwrite rather than append mode. I thought I had fixed that bug. Please could you create a ticket to remind me to investigate it? > 3) the following error message (see last row) seems to be incorrect: > "Can't open a file -- attempted to load a non-existant config file or bad > file referenced within?" when the file is really locked by another process: > this has nothing to do with a config file: It does end with a question mark. This is just a suggestion, and while the suggestion might not be perfect, sometimes it will be right. The real error message, cannot open a file, is still correct. Cheers, Chris. -- _ ___ __ _ / __/ / ,__(_)_ | Chris Wilson <0000 at qwirx.com> - Cambs UK | / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Perl/SQL/HTML Developer | \ _/_/_/_//_/___/ | We are GNU-free your mind-and your software | From achim+box at qustodium.net Thu Apr 8 16:03:10 2010 From: achim+box at qustodium.net (Achim) Date: Thu, 08 Apr 2010 17:03:10 +0200 Subject: [Box Backup] Logging In-Reply-To: References: <01c375e6352797b3ab14e3f524ce6590@localhost> Message-ID: Hello Chris: Thanks for the fast response! On Thu, 8 Apr 2010 16:25:18 +0200 (CEST), Chris Wilson wrote: > It is configurable: set NotifyAlways = yes if you want it. > >> "bang on if the user cannot be bothered to fix the problem". > > I think I agree that it should be on by default. Thanks for reminding me of the configuration option. I had seen it before but thought that I remembered some inconvenience with too many "false positives" (i.e., many mails with "everything is fine"), but of course that depends only on how the notify script handles the notifications. >> However, bbackupd.log is emptied at random intervals. > Could you create a ticket to remind me to investigate it? Done, >> 3) the following error message (see last row) seems to be incorrect: >> "Can't open a file -- attempted to load a non-existant config file or bad >> file referenced within?" when the file is really locked by another >> process: >> this has nothing to do with a config file: > > It does end with a question mark. This is just a suggestion, and while the > suggestion might not be perfect, sometimes it will be right. The real > error message, cannot open a file, is still correct. This is only a minor thing that I just came across, not really worth any of your attention. BUT the original Windows error (ERROR_SHARING_VIOLATION 32 (0x20), "The process cannot access the file because it is being used by another process.") is a lot more helpful, because the BB message actually points in the wrong direction. Perhaps there is a way to improve the estimated guess that BB ventures? Thanks again, Achim From joseyluis at gmail.com Sun Apr 11 02:32:49 2010 From: joseyluis at gmail.com (Jose Luis Rodriguez Garcia) Date: Sun, 11 Apr 2010 03:32:49 +0200 Subject: [Box Backup] Patch for raidfile test on NetBSD 4/5 In-Reply-To: References: Message-ID: On Tue, Apr 6, 2010 at 10:55 PM, Chris Wilson wrote: > Hi Jose, >..... > Thanks for the patch! Please do post it on our bug tracker to ensure that it > doesn't get forgotten. I will need to review it to ensure that it meets our > coding style and doesn't break other platforms. Will you continue to test > future versions of Box Backup on NetBSD 4/5? Would you mind being the > official (or unofficial) maintainer of support for this platform? Not problem, I can be the maintener of support for this platform. The only problem is that I can't dedicate much time to boxbackup. What is the difference between official or unofficial maintainer? > >> The problem is that in NetBSD 4/5 the readdir and openddir functions >> are called in libc: __readdir30 and __closedir30. >> readdir and closedir exist for compatibility with NetBSD 3 and before. >> >> There was a change in the ABI between the 3 and 4 version for enable >> 64 bits types. > > Perhaps we need to use the 64-bit versions instead? I guess that these, and > not the 32-bit versions, are called by the standard readdir() and closedir() > functions in C code? > This was a change for use inode of 64 bits in kernel instead of inodes of 32 bits in kernel. The functions continue to be the same when you code a program. 1- In the C library the old readdir and opendir are called: readdir and openddir 2- In the C library the new readdir and opendir are called __readdir30 and __opendir30 When you compile, a new program the dirent.h header rename the functions readdir/opendir to __readir30/__opendir30 It is for maintain compatibility binary. (NetBSD hasn't ever broken binary compatibility) readdir and opendir are the functions that are changed (I wrote closedir in my e-mail. It was closedir not opendir). From joseyluis at gmail.com Sun Apr 11 03:09:56 2010 From: joseyluis at gmail.com (Jose Luis Rodriguez Garcia) Date: Sun, 11 Apr 2010 04:09:56 +0200 Subject: [Box Backup] Example of testsuite output Message-ID: Could somebody send me the output of a correct execution of the testsuite: runtest.pl ALL I not sure that all error messages that I see are normal, altough the execution of runtest.pl ALL on my pc indicates that it have passed all the tests From brendon at netcal.com Mon Apr 12 07:43:49 2010 From: brendon at netcal.com (Brendon Baumgartner) Date: Sun, 11 Apr 2010 23:43:49 -0700 Subject: [Box Backup] Logging In-Reply-To: References: <01c375e6352797b3ab14e3f524ce6590@localhost> Message-ID: <5DB1D2D8020D4449AF3111C784529E6271890A@jeep.netcal.l> I was reading this logging thread. Would it be possible to have an option to reset the windows/linux log file once it reaches a specified size defined in the conf file? I have found 1gb logfiles on a few windows machines and as far as I can tell, there isn't a way to manage these files. Thanks -Brendon From chris at qwirx.com Mon Apr 12 10:19:35 2010 From: chris at qwirx.com (Chris Wilson) Date: Mon, 12 Apr 2010 11:19:35 +0200 (CEST) Subject: [Box Backup] Logging In-Reply-To: <5DB1D2D8020D4449AF3111C784529E6271890A@jeep.netcal.l> References: <01c375e6352797b3ab14e3f524ce6590@localhost> <5DB1D2D8020D4449AF3111C784529E6271890A@jeep.netcal.l> Message-ID: Hi Brendon, On Sun, 11 Apr 2010, Brendon Baumgartner wrote: > I was reading this logging thread. Would it be possible to have an > option to reset the windows/linux log file once it reaches a specified > size defined in the conf file? > > I have found 1gb logfiles on a few windows machines and as far as I can > tell, there isn't a way to manage these files. I'd prefer not to add such an option at the moment. I'd rather that this feature (LogFile option) not grow to infinite complexity. It's intended for two purposes: * Debugging, under close manual supervision * Parsing by the NotifyScript to produce backup reports, followed by deletion. Is it possible for you to use the Windows Event log for general-purpose logging and does that solve the problem for you? Cheers, Chris. -- _ ___ __ _ / __/ / ,__(_)_ | Chris Wilson <0000 at qwirx.com> - Cambs UK | / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Perl/SQL/HTML Developer | \ _/_/_/_//_/___/ | We are GNU-free your mind-and your software | From brendon at netcal.com Mon Apr 12 18:43:15 2010 From: brendon at netcal.com (Brendon Baumgartner) Date: Mon, 12 Apr 2010 10:43:15 -0700 Subject: [Box Backup] Logging In-Reply-To: References: <01c375e6352797b3ab14e3f524ce6590@localhost><5DB1D2D8020D4449AF3111C784529E6271890A@jeep.netcal.l> Message-ID: <5DB1D2D8020D4449AF3111C784529E6271890E@jeep.netcal.l> > On Sun, 11 Apr 2010, Brendon Baumgartner wrote: > > > I was reading this logging thread. Would it be possible to have an > > option to reset the windows/linux log file once it reaches a > > specified size defined in the conf file? > > > > I have found 1gb logfiles on a few windows machines and as far as I > > tell, there isn't a way to manage these files. > > I'd prefer not to add such an option at the moment. I'd rather that > this > feature (LogFile option) not grow to infinite complexity. It's intended > for two purposes: > > * Debugging, under close manual supervision > * Parsing by the NotifyScript to produce backup reports, followed by > deletion. > > Is it possible for you to use the Windows Event log for general-purpose > logging and does that solve the problem for you? I have used the event logs so far, but the agent is so chatty, it's quite disruptive to the administrator if they want to browse the eventlogs. Also, if the agent did one backup and then stopped at night, then it would probably be okay, but since it's constantly running, the eventlogs basically end being boxbackup logs because 90% of the line items are from boxbackup. -Brendon From james at netinertia.co.uk Mon Apr 12 19:54:33 2010 From: james at netinertia.co.uk (James O'Gorman) Date: Mon, 12 Apr 2010 19:54:33 +0100 Subject: [Box Backup] Logging In-Reply-To: <5DB1D2D8020D4449AF3111C784529E6271890E@jeep.netcal.l> References: <01c375e6352797b3ab14e3f524ce6590@localhost><5DB1D2D8020D4449AF3111C784529E6271890A@jeep.netcal.l> <5DB1D2D8020D4449AF3111C784529E6271890E@jeep.netcal.l> Message-ID: On 12 Apr 2010, at 18:43, Brendon Baumgartner wrote: > I have used the event logs so far, but the agent is so chatty, it's > quite disruptive to the administrator if they want to browse the > eventlogs. Also, if the agent did one backup and then stopped at night, > then it would probably be okay, but since it's constantly running, the > eventlogs basically end being boxbackup logs because 90% of the line > items are from boxbackup. To be honest this is a system administration issue and not an application issue. The eventlog viewer does let you filter out certain events, or filter on certain events. I frequently see the Wintel administrators at my place filter on certain codes to find account lockout events. They have also set up a job to export the eventlogs from all Domain Controllers and dump them into a SQL Server database. James From brendon at netcal.com Mon Apr 12 23:01:10 2010 From: brendon at netcal.com (Brendon Baumgartner) Date: Mon, 12 Apr 2010 15:01:10 -0700 Subject: [Box Backup] Logging In-Reply-To: References: <01c375e6352797b3ab14e3f524ce6590@localhost><5DB1D2D8020D4449AF3111C784529E6271890A@jeep.netcal.l><5DB1D2D8020D4449AF3111C784529E6271890E@jeep.netcal.l> Message-ID: <5DB1D2D8020D4449AF3111C784529E62718918@jeep.netcal.l> > On 12 Apr 2010, at 18:43, Brendon Baumgartner wrote: > > I have used the event logs so far, but the agent is so chatty, it's > > quite disruptive to the administrator if they want to browse the > > eventlogs. Also, if the agent did one backup and then stopped at > night, > > then it would probably be okay, but since it's constantly running, > the > > eventlogs basically end being boxbackup logs because 90% of the line > > items are from boxbackup. > > To be honest this is a system administration issue and not an > application issue. > > The eventlog viewer does let you filter out certain events, or filter > on certain events. I frequently see the Wintel administrators at my > place filter on certain codes to find account lockout events. They have > also set up a job to export the eventlogs from all Domain Controllers > and dump them into a SQL Server database. I disagree. If this was a linux system, then I would agree. There is no way with MS to show everything except Boxbackup entries. In addition, Windows applications don't fill up the eventlogs by default when the system is operating normally. Finally, exporting eventlogs to some management systems doesn't make sense for people with one or a few servers. -Brendon From achim+box at qustodium.net Mon Apr 12 23:42:49 2010 From: achim+box at qustodium.net (Achim) Date: Tue, 13 Apr 2010 00:42:49 +0200 Subject: [Box Backup] VSS Message-ID: <4BC3A1E9.5030406@qustodium.net> Hello Chris: Just tonight I came across Hobocopy [1], and I saw that you had left some comments about 4 years ago with respect to VSS. Given that the software uses the "MIT license" which "is also GPL-compatible, meaning that the GPL permits combination and redistribution with software that uses the MIT License" [2], would it be possible to revive your old work and "simply" copy/paste the code from Hobocopy, if all the abstraction layers from Bacula are too complicated? Thanks, Achim [1] [2] , From joseyluis at gmail.com Fri Apr 16 20:16:36 2010 From: joseyluis at gmail.com (Jose Luis Rodriguez Garcia) Date: Fri, 16 Apr 2010 21:16:36 +0200 Subject: [Box Backup] What are the minor problems of NetBSD with symlinks? Message-ID: What are the minor problems of NetBSD with symlinks? The main page: http://www.boxbackup.org/ states: NetBSD (minor issues with dates on symlinks) Could somebody provide details about these minor issues. I have checked the archives of boxbackup and I didn't find any reference to it, or any trac case. From lchalupa at seelink.org Sat Apr 24 17:31:58 2010 From: lchalupa at seelink.org (lchalupa at seelink.org) Date: Sat, 24 Apr 2010 09:31:58 -0700 Subject: [Box Backup] Introduction to BoxBackup Message-ID: <20100424093158.24918cff8389962577858d05cfdce4a0.c9edd383ed.wbe@email05.secureserver.net> An HTML attachment was scrubbed... URL: From nn4l at yahoo.de Mon Apr 26 21:02:33 2010 From: nn4l at yahoo.de (Arne Schirmacher) Date: Mon, 26 Apr 2010 20:02:33 +0000 (GMT) Subject: [Box Backup] restore to particular date and time? Message-ID: <828414.44969.qm@web26001.mail.ukl.yahoo.com> Is it possible to restore a directory to a particular date in the past, so that the versions of all files are not newer than that date? I can't find anything related in the documentation. I could of course restore the corresponding deleted version of a file but if there are lots of files this task is not manageable. From chris at qwirx.com Mon Apr 26 21:08:13 2010 From: chris at qwirx.com (Chris Wilson) Date: Mon, 26 Apr 2010 22:08:13 +0200 (CEST) Subject: [Box Backup] restore to particular date and time? In-Reply-To: <828414.44969.qm@web26001.mail.ukl.yahoo.com> References: <828414.44969.qm@web26001.mail.ukl.yahoo.com> Message-ID: Hi Arne, On Mon, 26 Apr 2010, Arne Schirmacher wrote: > Is it possible to restore a directory to a particular date in the past, > so that the versions of all files are not newer than that date? I can't > find anything related in the documentation. > > I could of course restore the corresponding deleted version of a file > but if there are lots of files this task is not manageable. I'm afraid there's no easy reliable way at the moment. See this feature request: http://www.boxbackup.org/trac/wiki/FeatureRequests#VersionedRepository There is a "kind of" restore to point in time feature in Boxi, to the best approximation that I could do with the current store format, but Boxi is not complete and there's absolutely no guarantee that it will restore your data or not bite your dog. Cheers, Chris. -- _ ___ __ _ / __/ / ,__(_)_ | Chris Wilson <0000 at qwirx.com> - Cambs UK | / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Perl/SQL/HTML Developer | \ _/_/_/_//_/___/ | We are GNU-free your mind-and your software | From brendon at netcal.com Wed Apr 28 00:01:32 2010 From: brendon at netcal.com (Brendon Baumgartner) Date: Tue, 27 Apr 2010 16:01:32 -0700 Subject: [Box Backup] Restoring from raw key file Message-ID: <5DB1D2D8020D4449AF3111C784529E62718A57@jeep.netcal.l> What are the basic steps for restoring from a raw key file? I have the raw key, but I'm not sure yet what to do with it. I'm hoping someone can help and speed my recovery up. I obviously have a project ahead of me :) -Brendon -------------- next part -------------- An HTML attachment was scrubbed... URL: From julien at nozav.org Wed Apr 28 11:08:48 2010 From: julien at nozav.org (Julien Barnier) Date: Wed, 28 Apr 2010 12:08:48 +0200 Subject: [Box Backup] Backup of an ecryptfs encrypted home Message-ID: <87sk6f3n4f.fsf@ens-lyon.fr> Hi, I'm a happy boxbackup user from several months now, and as such I first wanted to thank you for all your great work on it. And now the problem :-) I recently changed my linux box and setup an encrypted home directory with ecryptfs. So when boxbackup starts during the boot process, my home is still encrypted and only contains a .Private directory and a Readme.txt file. So Boxbackup immediately flagged all my files on the backup server as 'deleted'. When I log in, my home is mounted unencrypted, and then boxbackup begins to upload *all* of my home directory files again, as the previous ones have been marked as deleted. Do you think of a way to prevent this ? Is there a way to tell the boxbackup client to delay its files scan or not to start it as long as the home is not mounted ? Otherwise I thought of starting boxbackup client from my .bashrc or my .xsession, but it's far from being "clean"... Thanks a lot in advance, -- Julien From chris at qwirx.com Wed Apr 28 11:24:51 2010 From: chris at qwirx.com (Chris Wilson) Date: Wed, 28 Apr 2010 12:24:51 +0200 (CEST) Subject: [Box Backup] Backup of an ecryptfs encrypted home In-Reply-To: <87sk6f3n4f.fsf@ens-lyon.fr> References: <87sk6f3n4f.fsf@ens-lyon.fr> Message-ID: Hi Julien, On Wed, 28 Apr 2010, Julien Barnier wrote: > I recently changed my linux box and setup an encrypted home directory > with ecryptfs. So when boxbackup starts during the boot process, my home > is still encrypted and only contains a .Private directory and a > Readme.txt file. So Boxbackup immediately flagged all my files on the > backup server as 'deleted'. When I log in, my home is mounted > unencrypted, and then boxbackup begins to upload *all* of my home > directory files again, as the previous ones have been marked as deleted. > > Do you think of a way to prevent this ? Is there a way to tell the > boxbackup client to delay its files scan or not to start it as long as > the home is not mounted ? I'd suggest using the SyncAllowScript option to disable backups while your home directory is not mounted, by telling bbackupd to try again later, something like this: #!/bin/sh if mount | grep "/home/julien type ecryptfs"; then echo "now" else echo "60" fi Cheers, Chris. -- _ ___ __ _ / __/ / ,__(_)_ | Chris Wilson <0000 at qwirx.com> - Cambs UK | / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Perl/SQL/HTML Developer | \ _/_/_/_//_/___/ | We are GNU-free your mind-and your software | From julien at nozav.org Wed Apr 28 13:40:08 2010 From: julien at nozav.org (Julien Barnier) Date: Wed, 28 Apr 2010 14:40:08 +0200 Subject: [Box Backup] Backup of an ecryptfs encrypted home In-Reply-To: (Chris Wilson's message of "Wed, 28 Apr 2010 12:24:51 +0200 (CEST)") References: <87sk6f3n4f.fsf@ens-lyon.fr> Message-ID: <87mxwn3g47.fsf@ens-lyon.fr> Hi Chris, > I'd suggest using the SyncAllowScript option to disable backups while > your home directory is not mounted, by telling bbackupd to try again > later, something like this: Of course, I'm quite sorry to have missed this one. I didn't know the existence of this feature, but I could have learnt it by reading my bbackupd.conf more slowly :-) I just added a small /dev/null redirect to your script to suppress the grep output : ,---- | #!/bin/sh | | if mount | grep "/home/julien type ecryptfs" > /dev/null; then | echo "now" | else | echo "60" | fi `---- Thanks a lot for your help ! Julien From chris at qwirx.com Wed Apr 28 13:54:28 2010 From: chris at qwirx.com (Chris Wilson) Date: Wed, 28 Apr 2010 14:54:28 +0200 (CEST) Subject: [Box Backup] Backup of an ecryptfs encrypted home In-Reply-To: <87mxwn3g47.fsf@ens-lyon.fr> References: <87sk6f3n4f.fsf@ens-lyon.fr> <87mxwn3g47.fsf@ens-lyon.fr> Message-ID: Hi Julien, On Wed, 28 Apr 2010, Julien Barnier wrote: > I just added a small /dev/null redirect to your script to suppress the > grep output : Adding the -q option to grep removes the need for this, and is slightly more efficient :) Sorry I missed that. Cheers, Chris. -- _ ___ __ _ / __/ / ,__(_)_ | Chris Wilson <0000 at qwirx.com> - Cambs UK | / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Perl/SQL/HTML Developer | \ _/_/_/_//_/___/ | We are GNU-free your mind-and your software | From brendon at netcal.com Wed Apr 28 15:59:05 2010 From: brendon at netcal.com (Brendon Baumgartner) Date: Wed, 28 Apr 2010 07:59:05 -0700 Subject: [Box Backup] Restoring from raw key file In-Reply-To: <5DB1D2D8020D4449AF3111C784529E62718A57@jeep.netcal.l> References: <5DB1D2D8020D4449AF3111C784529E62718A57@jeep.netcal.l> Message-ID: <5DB1D2D8020D4449AF3111C784529E62718A58@jeep.netcal.l> Never mind. I was over-complicating things. I was assuming the Raw key file had something to do with the open-ssl encryption, which apparently it doesn't. -Brendon From: boxbackup-bounces at boxbackup.org [mailto:boxbackup-bounces at boxbackup.org] On Behalf Of Brendon Baumgartner Sent: Tuesday, April 27, 2010 4:02 PM To: boxbackup at boxbackup.org Subject: [Box Backup] Restoring from raw key file What are the basic steps for restoring from a raw key file? I have the raw key, but I'm not sure yet what to do with it. I'm hoping someone can help and speed my recovery up. I obviously have a project ahead of me :) -Brendon -------------- next part -------------- An HTML attachment was scrubbed... URL: From lchalupa at seelink.org Fri Apr 30 17:25:34 2010 From: lchalupa at seelink.org (lchalupa at seelink.org) Date: Fri, 30 Apr 2010 09:25:34 -0700 Subject: [Box Backup] rpmbuild for Fedora 12 does not work Message-ID: <20100430092534.24918cff8389962577858d05cfdce4a0.ef953ee315.wbe@email05.secureserver.net> An HTML attachment was scrubbed... URL: From chris at qwirx.com Fri Apr 30 18:15:42 2010 From: chris at qwirx.com (Chris Wilson) Date: Fri, 30 Apr 2010 19:15:42 +0200 (CEST) Subject: [Box Backup] rpmbuild for Fedora 12 does not work In-Reply-To: <20100430092534.24918cff8389962577858d05cfdce4a0.ef953ee315.wbe@email05.secureserver.net> References: <20100430092534.24918cff8389962577858d05cfdce4a0.ef953ee315.wbe@email05.secureserver.net> Message-ID: Hi Lee, On Fri, 30 Apr 2010, lchalupa at seelink.org wrote: > Hello.? Trying to install boxbackup. Getting error. > > $ rpmbuild -ta boxbackup-0.11rc7.tgz > > This is the error I'm getting.... > > + install -m 644 VERSION.txt/root/rpmbuild/BUILDROOT/boxbackup-0.11rc7-1.fc12.i386/usr/share/doc/boxbackup-0 > .11rc7 > + install -m 644 ExceptionCodes.txt/root/rpmbuild/BUILDROOT/boxbackup-0.11rc7-1.fc12.i386/usr/share/doc/boxbackup-0 > .11rc7 > + install -m 644 LICENSE.txt/root/rpmbuild/BUILDROOT/boxbackup-0.11rc7-1.fc12.i386/usr/share/doc/boxbackup-0 > .11rc7 > install: cannot stat `LICENSE.txt': No such file or directory > error: Bad exit status from /var/tmp/rpm-tmp.Fwn0sB (%install) > > I'm stuck. Thanks for reporting the problem. It will be fixed in the next release candidate and 0.11 final. In the mean time, you can untar the distribution tarball, cd into boxbackup-0.11rc7, edit contrib/rpm/boxbackup.spec, and find the following lines at around line 108: install -m 644 LICENSE.txt \ $RPM_BUILD_ROOT%{_docdir}/%{ident} and replace them with the following four lines: install -m 644 LICENSE-GPL.txt \ $RPM_BUILD_ROOT%{_docdir}/%{ident} install -m 644 LICENSE-BSD.txt \ $RPM_BUILD_ROOT%{_docdir}/%{ident} This same change will be in the next release. Thanks again. Cheers, Chris. -- _ ___ __ _ / __/ / ,__(_)_ | Chris Wilson <0000 at qwirx.com> - Cambs UK | / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Perl/SQL/HTML Developer | \ _/_/_/_//_/___/ | We are GNU-free your mind-and your software |